What Is a Dictatorship? Power Without a Way to Remove It

Political scientists generally classify regimes by whether power can change hands through a process the incumbent does not control. On that test the diagnostic features are consistent:

• •No genuinely competitive elections, meaning either none at all or ones whose outcome is not in doubt because the opposition is restricted, the media controlled or the count unreliable
• •No independent judiciary able to rule against the executive and have the ruling obeyed
• •Restricted freedom of association, assembly and the press, so that organised opposition cannot form
• •Security services accountable to the ruler rather than to law, with detention, surveillance and violence available as routine instruments
• •No fixed limit on tenure, or limits that are amended whenever they approach
• •Crucially, the presence of elections, courts, parliaments and constitutions does not disqualify a regime, since most modern authoritarian states maintain all of them in forms that cannot produce an unwanted outcome

The category covers arrangements that work very differently. Personalist regimes concentrate power in one individual, with institutions subordinated to their preferences and succession unresolved, which makes them volatile and prone to collapse at the ruler's death. Single-party regimes locate power in an organisation that selects and can remove leaders, which historically makes them the most durable form because the party manages succession and distributes patronage. Military regimes are typically the least stable and the shortest-lived, since officers generally want to return to the barracks and the institution itself has interests that diverge from ruling. Monarchies persist where a royal family manages internal succession and distributes resources. Totalitarian regimes, in Arendt's sense, are a distinct and rare subtype attempting comprehensive control of society, ideology and private life rather than merely monopolising power, which is why the label applies to very few states.

The most significant change of the last thirty years is that the classic form has largely been replaced by something less obvious. Competitive authoritarianism, described by Steven Levitsky and Lucan Way, maintains genuinely contested elections that the opposition cannot win because the playing field is tilted: state media favours the incumbent, state resources fund the campaign, tax authorities and regulators harass opponents and their business backers, courts rule reliably, and violence is used selectively rather than systematically. The mechanisms are legal in form, which makes them difficult to denounce and easy to defend as ordinary governance. The pattern of erosion is well documented and remarkably consistent: capture of the courts, capture or intimidation of media ownership, redrawing of electoral rules, restriction of civil society funding as foreign interference, and attacks on the legitimacy of any institution that produces an adverse finding. The process is gradual and each individual step is defensible, which is precisely the design.

Durability rests on a small set of things: control of the security forces, which is why loyalty is bought and rival forces are created as counterweights; distribution of resources to a coalition whose support is necessary, which is the core of selectorate theory as set out by Bruce Bueno de Mesquita and colleagues; and the prevention of coordination among opponents, since opposition requires people to know that others share their view, which censorship prevents more effectively than it prevents individual belief. Failure comes through a corresponding set: a succession crisis in a personalist regime, an economic collapse that removes the resources holding the coalition together, a split within the elite, a security force that refuses to fire, or a moment of public coordination in which the private opposition becomes visible all at once, which is why the fall of such regimes so often looks sudden. Timur Kuran's analysis of preference falsification explains the suddenness directly: when everyone conceals their real view, nobody knows how many share it, and a small revelation can cascade.

Several projects attempt to measure this systematically, including Freedom House, the Varieties of Democracy institute at Gothenburg and the Polity dataset, and they use different definitions and produce broadly similar trends. Their common finding for the last fifteen years is a decline in the number of people living under democratic government, driven less by coups than by erosion within states that hold elections, and with the largest movements in populous middle-income countries rather than in the states that were never democratic. The measurement itself is contested, since coding a regime requires judgement and the indices have been criticised for conceptual stretching and for scoring that reflects the values of those constructing them. What is not seriously disputed is the mechanism by which erosion occurs, which has been documented closely enough in enough cases that the sequence is recognisable while it is happening.

The Roman office the word comes from was temporary, legal and limited to six months, and the modern meaning is defined by the absence of any mechanism the incumbent does not control for removing them. Elections, courts and constitutions are usually present and unable to produce an unwanted outcome. The varieties differ in durability, with single-party regimes the most stable and military ones the least, and the dominant modern form tilts a genuinely contested field using legal instruments. Regimes last through control of security forces and distribution to a necessary coalition, and fall suddenly when concealed opposition becomes visible at once.